Georgia court grants OpenAI summary judgment in first AI defamation case

A Georgia state court granted OpenAI summary judgment in Walters v. OpenAI, the first defamation lawsuit brought over a chatbot’s hallucinated output to reach a ruling. Judge Tracie Cason of Gwinnett County State Court dismissed the case on three independent grounds, closing off what plaintiffs’ lawyers had watched as a test of whether existing defamation law could reach AI-generated falsehoods at all.

The claim arose in 2023, when a journalist asked ChatGPT to summarise a real federal lawsuit and the model incorrectly described it as accusing radio host Mark Walters of embezzlement — an allegation the actual complaint did not contain. The journalist checked the underlying document, recognised the error and never published the false summary, but Walters sued OpenAI directly over the hallucinated output itself.

The court found, first, that no reasonable user could have understood the output as a factual, defamatory statement about Walters, given OpenAI’s prominent warnings that ChatGPT can produce inaccurate information and the surrounding context of the journalist’s own query. Second, it held Walters had not shown OpenAI met the required fault standard — neither actual malice nor negligence — noting the company’s ongoing efforts to reduce hallucination rates. Third, and independently sufficient on its own, the court found Walters had not established any damages: he testified at deposition that he had not been harmed and was not seeking damages in the case.

Because the ruling turned on the specific facts — an unpublished hallucination, prominent warnings and an admission of no harm — lawyers on both sides cautioned it did not resolve the broader question of AI-developer liability for hallucinated defamatory content, but it gave the first judicial framework for how such claims would be tested.